Mistral AI API

Mistral AI API provides Mistral Large / Mistral Small / Mistral NeMo / Codestral — European open-source / hybrid LLMs, the European answer to OpenAI.

Visit site ↗Documentation ↗Health checked 9h ago

Use it when

Partial models fully open-source (Mistral 7B, Mixtral 8x7B, Mistral NeMo)

Watch for

Strongest model Mistral Large still lags GPT-4 / Claude 3.5

First check

Sign up at console.mistral.ai for API key. POST https://api.mistral.ai/v1/chat/completions with model: "mistral-large-latest" + messages.

Auth

api_key

CORS

HTTPS

Yes

Signup

Latency

609 ms

Protocol

REST

Pricing

paid

Uptime · 30-day window

Probes: 1Uptime: 100%Avg latency: 609ms

About this API

Mistral AI is a French LLM company founded 2023 (founding team from Meta / DeepMind), became a unicorn in three months. Hailed as "European OpenAI" — actively supported by EU government as core to European AI sovereignty strategy. Product line: (1) Mistral 7B (open-sourced 2023, once best small open model); (2) Mixtral 8x7B and Mixtral 8x22B (mixture-of-experts open models, performance near GPT-3.5); (3) Mistral Large (flagship closed model, quality approaching but below GPT-4); (4) Codestral (code generation specialist); (5) Mistral NeMo (open-source 12B model partnered with NVIDIA). API is practical and modern — function calling, JSON mode, streaming all supported. European enterprises prefer Mistral because datacenters are in EU (GDPR-friendly). Less used by Chinese customers; mainly European market.

What you can build

1European compliance (data stays in EU)
2Function calling and code generation
3Mid-size model deployment on own GPUs (Mistral 7B/8x7B fully open-source)

Strengths & limitations

Strengths

Partial models fully open-source (Mistral 7B, Mixtral 8x7B, Mistral NeMo)
European datacenters (GDPR-friendly)
Mid-range pricing (Mistral Large $2/$6 per M)

Limitations

Strongest model Mistral Large still lags GPT-4 / Claude 3.5
API ecosystem weaker than OpenAI

Example request

Generic template — replace <endpoint> with the real path from the docs.

curl https://mistral.ai/<endpoint> \
  -H "Authorization: Bearer $API_KEY"
# Some providers use X-Api-Key instead — verify in the docs.

Getting started

Sign up at console.mistral.ai for API key. POST https://api.mistral.ai/v1/chat/completions with model: "mistral-large-latest" + messages.

FAQ

How do I use the open-source models?+

Mistral 7B etc. run locally with ollama / Hugging Face transformers. Mistral Large is closed-source API-only.

How is Chinese support?+

Chinese quality is average (lags China-domestic models like Qwen). For multilingual tasks, recommend Cohere or GPT-4.

Technical details

CORS: ?HTTPS: YesSignup: ?Open source: No

Auth type: api_key
Pricing: paid
Rate limit: free trial 1 RPM；付费 tier 提升到 100+ RPM
Protocols: REST
SDKs: python, typescript, javascript
Response time: 609 ms
Last health check: 5/12/2026, 7:37:53 AM